Their goal is for the GPT-5 update AI to feel more like a friend and less like a computerized encyclopaedia, especially after users complained the earlier version sounded too stiff. Some folks love the friendlier update, while others grumble they never asked for a buddy in a chatbot. This divide shows the tightrope OpenAI still walks when deciding between ease of use and pure, serious answers.
In the update announcement OpenAI talked about the little verbal tricks the AI now uses. Rather than sounding strictly formal, the bot might kick things off with, “Good question” or “Great start” before getting to the answer. Officially, they insist the new phrasings aren’t cookie-cutter compliments but cues to soften the AI’s edges. They even ran internal tests and the number of sappy compliments being dropped into replies is about the same as in the previous version.
Mixed Reactions from Users
Right after the update started going live, every platform you can name started buzzing with takes, and truth is not everybody is loving it. A good chunk of folks say they never signed up for the AI buddy experience. They want the assistant to fire back with facts, not to chit-chat like it’s hanging out for coffee. One person got straight to the point: “Stop pretending to be human. You’re a machine, sound like it.” Another added they’d rather have a helper close by, not a pen pal.
Some people are annoyed that OpenAI is spending time making ChatGPT sound more friendly instead of making it work better. “Please don’t make this a Pro feature,” one guy tweeted. “I just want it to answer my questions. If I wanted a real human to chat with I’d go outside.”
But not everyone was grumpy. Plenty of folks liked the new friendly vibe. They said it made the chat easier to use when they were looking for tips or brainstorming. A few people even suggested a neat idea: what if the app had a switch so you could choose between a straight shooter or a chatty buddy?

The Cool-Off on GPT-5 update
The so-so response to this little tweak is the latest tumble on the GPT-5 rollout, which dropped a couple of weeks ago. Everybody was super hyped, but now a lot of users say it’s a letdown. Some swear it’s worse at coding and giving in-depth answers than GPT-4. Others are mad that GPT-4 vanished from the free tier, so they’ve got to stick with a version they don’t like.
OpenAI is hustling to sort out the newest complaints. First, the team brought back the old models like GPT-4o, GPT-4.1, o4-mini, and o3, but they only gave access to Plus and Pro subscribers. After folks pushed back on GPT-5 always deciding the answer style itself, the team rolled out four clear response options:
- Auto lets the AI pick the best way to reply.
- Fast gives super-short answers.
- Thinking goes for the full, step-by-step reasoning.
- Thinking Mini tries to land between short and smart.
These changes prove OpenAI is still testing the best way for GPT-5 updateo to chat, and that they’re listening closely to the feedback.
The Bigger Question: What Do We Want from AI?
The buzz around whether GPT-5 should be “friendly” gets people wondering what we really want from AI in the first place. Do we want it to stick to the facts, no fluff, just the instructions we asked for? Or are we hoping for a little kindness, friendly nudges, and a chat that feels like talking to a pal who just wants us to do better?
Picture this: you ask ChatGPT how to solve a tricky math problem, write a line of code, or name a few dates from history. A cold, no-nonsense answer feels just right. But swap the question to “What career should I pick?” or “I’m feeling down, can you help?” and suddenly a warm tone can turn the chat from a tool to a teammate. The problem is that yesterday’s friendly answer can annoy the next person who just wants the facts.
That’s why a lot of tech people are betting that the AI of the future will come with customization built in. What if we could flip a switch to pick the voice we want—strict, chill, pep-rally cheerful, or straight shooting? The ability to dial the tone will probably be the key to keeping everyone happy and the chat feeling just right.
What’s Coming Next for GPT-5 update
Looking down the road, it feels like OpenAI won’t just slide into a standard, one-size-fits-all solution. The team’s dropped hints that more tweaks are rolling out in the next few weeks, which feels like a promise that GPT-5’s vibe and skills will keep changing. When you think about the past, it’s pretty likely we’ll one day get sliders and options for fine-tuning the personality on the fly, maybe with pre-set styles we can just click and go.
Meanwhile, the rollout showed that trying to make AI easy while cramming in tons of power is trickier than it seems. The first version of GPT-5 showed up with a stripped-down model, ditching the buffet of modes and styles. Even with the goal of a cleaner, simple launch, people pushed back, and the team had to walk that back and re-include a lot of the old options. The back-and-forth scrolls through the main issue: users want a single button easy mode, yet still expect detailed customization on the next screen. Balancing that is like trying to juggle bricks on a balance beam.
A Look Back and Ahead

OpenAI has been here before. Soon after the rollout of the new GPT-4o in early 2024, we saw the familiar wave of discontent. Many users found the latest update felt shakier than the earlier versions. The pattern—lots of buzz, quick push-back, and adjustments—has greeted almost every major model launch, reminding us how wide and picky the AI user crew is.
While the tale of GPT-5 is still being written, it already wears a softer, kinder façade. Still, the mixed reactions prove it’s no slam dunk; some users still aren’t ready for a model that mimics a more human tone. OpenAI is busy smoothing the technology, yet the tougher question hangs in the air: can one model be both the straight-shooting assistant some folks want and the chatty friend others enjoy?
If the team nails that tightrope, GPT-5 update just might turn skeptics into fans. Miss the mark, and the community could keep asking for the “quiet” versions that spoke in facts rather than small talk. Regardless, the argument over how AI “should” talk is just warming up—and that debate might steer human-robot conversations for a long while ahead.